选择groupby中的前3个元素 [英] Selecting top 3 elements within groupby

问题描述

鉴于以下按表分组的熊猫，我如何才能在每个CPUCore中获得Offline_RentetionAge的前3个index，并保持表的结构?

Given the following pandas group by table, how could I get the top 3 index for Offline_RentetionAge within each CPUCore, and keep the structure of the table?

对于上述结果，应该是

   CPUCore  Offline_RetetionAge  index
0        i7     183                4184
1               7                  1981
2               30                  471
3        i5     ..                ...

推荐答案

您可以使用

You can use GroupBy.head if values in index column are sorted:

df = df.groupby(level=0).head(3)

示例:

df = pd.DataFrame({'CPUCore':['i7'] * 4 + ['i5'] * 4,
                    'Offline_RetetionAge':['100','1','12','120','15','10','20','3'],
                   'index':[11,16,5,4,30,18,2,1]})
       .set_index(['CPUCore','Offline_RetetionAge'])

print (df)
                             index
CPUCore Offline_RetetionAge       
i7      100                     11
        1                       16
        12                       5
        120                      4
i5      15                      30
        10                      18
        20                       2
        3                        1

df = df.groupby(level=0).head(3)
print (df)
                             index
CPUCore Offline_RetetionAge       
i7      100                     11
        1                       16
        12                       5
i5      15                      30
        10                      18
        20                       2

这篇关于选择groupby中的前3个元素的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！